South China, especially Guangdong Province, has witnessed an alarming increase in both reported syphilis infections1,2 and sexually transmitted HIV infections (Fig. 1).3,4 Syphilis increases the risk of HIV acquisition and transmission, and patients with either sexually transmitted infection (STI) frequently attend the same STI clinics.5 Although the syphilis epidemic and sexually transmitted HIV epidemic have common high-risk groups and testing sites in South China,6 widespread syphilis testing has been successful although HIV testing has had poor uptake.7,8 An estimated 71% of those with HIV do not know their serostatus in China,9 and high levels of test refusal have been noted at STI clinics.10 HIV voluntary counseling and testing (VCT) sites in China have had problems attracting high-risk groups.7 Addressing these challenges, China issued draft provider-initiated testing and counseling guidelines and established provider-initiated HIV testing programs in 4 provinces.7
However, transforming a single HIV test into a comprehensive HIV screening program is no small task, complicated by several factors that limit routine HIV screening, particularly among the groups at higher risk of sexually transmitted HIV infection. Integration of HIV testing into routine clinical care has proven useful for expanding provider-initiated testing and counseling in a number of clinical settings.11 Normalizing HIV testing within the framework of existing routine screening tests may help decrease barriers to HIV testing at several levels (Fig. 2). At the individual level, high-risk individuals may be more willing to accept an HIV test if it is offered in tandem with syphilis testing as part of a routine sexual health screening experience. At the interpersonal/dyadic level, peer effects of those who accompany individuals to be tested could influence integrated test uptake. Finally, physician-level determinants related to linked syphilis/HIV training and stigma might affect integrated syphilis/HIV testing programs.
The highly related epidemiology, biology, and social context of syphilis and HIV infection suggest integrated testing programs in China. In addition, there have been very few analyses of HIV testing at STI clinics in China despite the expansion of these twin epidemics. The objective of this study was to analyze determinants of syphilis and HIV test uptake at STI clinics in China to establish a model for integrated syphilis/HIV screening.
Guangdong Province, South China, was chosen for this study because of recent increases in reported sexually transmitted HIV infection4 and syphilis.1 In 2007, the total burden of syphilis in China was 17.2 syphilis cases per 100,000 population with Guangdong Province reporting 31.6 cases per 100,000 population.12 Nearly half of Guangdong Province's primary syphilis cases are clustered in the more developed Pearl River Delta Region,13 a more developed region that has many counties reporting higher primary syphilis case rates.14 Like other provinces in China, Guangdong has a network of public STI clinics that reaches down to the township level and provides syphilis testing, HIV testing, and related clinical STI services according to national guidelines.1
The Plum Blossom Project was launched in September 2009 as part of a Guangdong province-wide effort to expand syphilis and HIV testing at public STI clinics. A collaboration of both local government and international agencies, the Plum Blossom Project ensured that free rapid syphilis and free rapid HIV testing were widely available at STI clinics. The Pearl River Delta region of Guangdong has a disproportionate burden of sexually transmitted HIV infection4 and syphilis.15 Three municipalities in the Pearl River Delta were chosen to implement this research based on laboratory capacity, availability of trained research assistants, and space to conduct the interview. Two public STI clinics in each of the 3 cities were selected using a probability proportional-to-size sampling method. Administrative data were used to estimate the number of public STI patients evaluated per day at each site. The total number of study participants at each site ranged from 336 to 628 over the course of the 5-month study. Each site had 4-14 STI physicians who evaluated patients.
The first research phase was a formative qualitative study to test survey items and determine local syphilis/HIV testing capacity, referral patterns, and the availability of free syphilis/HIV treatment and care services. All of these essential characteristics were present at local Guangdong public STI clinics, and more detailed results of the initial feasibility study are described elsewhere.16 All items from the physician survey were field tested among local Guangdong Province STI physicians, and a series of one-on-one interviews and focus groups helped refine the survey format.
The STI physician survey consisted of 3 domains—sociodemographic and training background information, clinical scenarios and vignettes, and stigma assessment. All physicians who completed the survey had a medical license and formal medical training according to national guidelines.17 Sociodemographic and training items were adapted from other published measures,18 whereas stigma items were validated among Chinese physicians in a previous study.19 All physicians were given the survey before Plum Blossom Project training, and survey completeness was checked by research assistants.
The STI patient survey consisted of ten items, including marital status, previous HIV testing, monthly income, age, whether or not they were accompanied or alone, education completed, date of survey, duration of local residency, occupation, and sex. STI patient survey items were previously validated in the China Family Health and Life Survey.20 All surveys were self-administered and checked by a research assistant for completeness. In addition, research assistants filled out a short section linking each patient to their respective physician and asking the patient about the following: whether the physician offered syphilis/HIV testing, whether the participant was willing to be tested for syphilis/HIV, whether the participant accepted syphilis/HIV testing, whether the participant returned for results, and the result of the rapid syphilis or rapid HIV test. The primary outcome of this study was test uptake defined by receipt of test results.
From September 2009 until January 2010, potential study subjects were identified by physicians, nurses, laboratory staff, and research assistants at selected STI clinics. All STI patients older than 17 years were eligible for participation, regardless of their interest in receiving syphilis or HIV testing. Patients who refused both HIV and syphilis testing had a 10-item survey completed by their physician; patients who accepted either HIV testing, syphilis testing, or both were asked to fill out the longer form. Participation in the survey was voluntary, and no incentives were given to physicians or patients to participate in the study. Those who agreed to participate in the study were given STI/HIV counseling after providing verbal informed consent. All patients received appropriate medical care for STI/HIV according to national guidelines. Venous samples were tested for HIV by ELISA (Acon, Hangzhou, China) and syphilis by an immunochromatographic rapid treponemal test (Wantai, Beijing, China). Both the rapid syphilis test and rapid HIV test were approved by China's State Food and Drug Administration and then underwent further local field evaluation. Test results for both of the rapid tests were available within 30 minutes. STI patients were told to wait for the results of their tests. Post-test counseling for syphilis/HIV testing was implemented according to standard STI clinic procedures as each of these clinics already diagnosed both syphilis and HIV infection.
All participants provided informed consent, and this research protocol was approved by the Medical Ethics Committee of Chinese Academy of Medical Sciences Institute of Dermatology (Nanjing, China), the University of North Carolina Public Health and Nursing IRB (Chapel Hill, NC), and the Partners Committee on Human Subjects Research (Boston, MA).
The main outcomes in this analysis were HIV and syphilis test uptake as defined by patient receipt of test results, both analyzed as dichotomous outcomes. Test uptake was defined as patient receipt of results because this is the most relevant outcome for operationalizing widespread HIV screening. Because only 10 survey items were available from both those who accepted and refused HIV testing, only these variables were analyzed in the multilevel models. Data were double entered into a database, and SPSS 17.0 (Chicago, IL) was used to perform all the nonmodeling statistical analyses. Kendall tau-b tests21 were used to estimate the probability of syphilis/HIV test uptake concordance.
Given our known clustering in outcomes and the potential for interactions between individual-level and physician-level predictors of HIV and syphilis test outcome, we used multilevel modeling to analyze the primary outcome.22,23 Because patients were allowed to enroll only once in the study, each patient was assigned to exactly 1 physician, and the data set was naturally hierarchically organized.
A 2-level model binomial logit model incorporated patient (level 1) and physician (level 2) variables. Two dyadic (marital status, accompaniment) variable were treated as level 1 variables because more detailed information about those who accompanied participants to the clinic was not available. Although city and clinic-level factors are potentially important determinants of HIV test uptake, we had an insufficient number of cities (n = 3) and clinics (n = 6) to create a level 3 variable. Instead, city was treated as a potential level 2 covariate of interest. The binomial logit 2-level model was of the form:
This is a binomial logistic multilevel model with random intercepts, and the binary response yij equals 1 if the individual i who saw physician j was tested for HIV. The variable Xij is a set of independent variables measured at the individual level. The intercept consists of a fixed component β 0 and a physician-specific component, the random effect u0j. The assumptions of homoskedasticity and independent residuals were empirically analyzed (see Figure, Supplemental Digital Content 1, http://links.lww.com/QAI/A165). Patient-physician interactions related to age (less than 10 years age difference compared with more than 10 years age difference) and sex (concordant versus discordant) were also assessed. The HIV test uptake model was constructed using MLwiN 2.20 (Bristol, United Kingdom), a software program used in multilevel modeling (http://www.cmm.bristol.ac.uk/MLwiN/), and Markov Chain Monte Carlo estimators.24,25 Missing values accounted for less than 5% of the outcome variables.
A total of 2069 STI patients were approached by 62 physicians about taking part in this study. Eight STI patients were younger than 18 years old and not eligible. Among those eligible, a total of 271 (13.1%) participants refused both tests and their physicians filled out a 10-item refuser form. The remainder of the patients (1790, 86.9%) filled out a more complete survey instrument. Among all those eligible (n = 2061), 1681 (81.6%) had syphilis test uptake and 1673 (81.2%) had HIV test uptake. All of the patients who received an HIV test also received a syphilis test, but 8 STI patients only received a syphilis test. This was the first syphilis or HIV test for the large majority (1388, 77.7%) of participants. More detailed information about syphilis testing compared with HIV testing can be found in Table 1. The most highly correlated measures were syphilis/HIV tests performed (correlation coefficient 0.957, P value < 0.001) and willing to be tested for syphilis/HIV (correlation coefficient = 0.939, P < 0.001). Having a doctor offer syphilis testing was associated with having the doctor offer HIV testing (P < 0.001); patient returning for syphilis test results was also associated with the patient returning for HIV test results (P < 0.001).
Several individual and interpersonal factors were significantly associated with HIV test refusal in the most parsimonious multilevel model (Table 2). No physician level factors significantly improved the most parsimonious model. The final model predicting HIV test refusal included the following factors: being married, having a previous HIV test, being unaccompanied, and participating in the last 2 months of the study (see Table, Supplemental Digital Content 2, http://links.lww.com/QAI/A166). Being married and having prior HIV testing were the strongest predictors of HIV test refusal. Married participants had an HIV uptake of 80.3% compared with unmarried participants who had an HIV uptake of 98.0% [adjusted odds ratio (aOR): 14.59]. Participants who received HIV testing in the past had an HIV uptake of 67.3% compared with those who never received HIV testing who had an HIV test uptake of 95.7% (aOR: 6.49). Among those who reported HIV testing in the past, 213 (37.6%) reported testing in the past 6 months, 208 (36.7%) reported testing between 6 months and 12 months ago, and 146 (25.7%) reported being tested between 1 and 5 years ago. Unaccompanied participants had an HIV test uptake of 84.5% compared with those who were accompanied (significant other, spouse, or other) who had an HIV test uptake of 91.2% (aOR: 1.82). Participants who enrolled in the last 2 months of the study had a mean HIV test uptake of 83.7% compared with those who enrolled in the first 3 months of the study who had an HIV test uptake of 88.8% (aOR: 1.46). Age, local residency, occupation, education, and sex did not improve the most parsimonious model. Given the high concordance between syphilis test uptake and HIV test uptake, the factors associated with HIV test uptake were also included in the most parsimonious model predicting syphilis test uptake.
At level 2, physician-level variance accounted for 28.4% of the total variance, assuming the patient-level variance corresponds to that of a standard logistic distribution. None of the physician-level predictors improved the model as follows: physician city, physician age, physician sex, physician medical degree, physician HIV training, physician HIV stigma measured using a validated survey, patients evaluated per day by the physician. Neither of the following patient-physician interaction variables substantially improved the model as follows: sex of patient/physician, age of patient/physician.
This study reveals an opportunity for scaling up integrated syphilis/HIV testing at STI clinics in China. About 77% of this patient sample had never had a syphilis test or HIV test before this visit, suggesting that STI clinic-based syphilis/HIV screening is uncommon. In the context of China, where 59% of HIV infections are thought to be sexually transmitted,7 scaling up STI clinic-based HIV testing represents a feasible logical extension of other testing efforts. To our best knowledge, this study is the first to analyze multilevel predictors of HIV test uptake in a clinical setting. Although multilevel modeling has been widely applied in the fields of education and sociology, there have been far fewer studies in medicine and sexual health in particular.26,27 Multilevel modeling presents advantages in describing clustered observations because it provides more accurate standard errors, accounts for patient-physician interactions, and permits analysis of physician-level predictors. Previous studies of HIV testing have either ignored clustering of observations or analyzed them using generalized estimating equations.
The high level of integrated syphilis/HIV test uptake, 81%, found at Chinese STI clinics represents marked improvement compared with lone HIV testing at STI clinics and the current system of VCT. One study of isolated HIV testing at STI clinics did not measure HIV test uptake but found that approximately 60% of Chinese STI patients accepted a lone HIV test, compared with 94.5% of STI patients in our sample accepting an HIV test.10 Studies from the general population,8 sex workers,28 and migrants29 have noted infrequent attendance and poor HIV test uptake at government-designated VCT centers. The high level of integrated syphilis/HIV test acceptance observed corresponds to high levels of integrated HIV test acceptance reported in Chinese HIV/TB integrated programs (99%)30 and antenatal care/HIV integrated programs (84%).31 The prenatal care system in China has launched programs to expand integrated HIV/HBV/syphilis testing31 that could provide useful guidance for HIV test integration more generally. Our research and these related integrated HIV testing programs highlight the benefits of provider-initiated HIV testing in China.
The association between being married and refusing HIV testing has been noted in quantitative HIV uptake studies from Malawi32 and the United States.33 Married individuals may decline HIV testing because a positive test result could have consequences for their spouse and marriage. In China, a large number of Chinese men who have sex with men are married,34 and this may also decrease HIV test uptake in this population. The association between prior HIV testing and test refusal has also been noted elsewhere,32 but may be particularly important in China where few education programs have focused on promoting HIV testing and a 1-time massive HIV testing program was implemented in 2003-2004 with minimal counseling.35 There have been no studies on HIV test uptake as it relates to accompaniment. Other studies of couples VCT have been encouraging,36 and our finding that those who are accompanied have higher HIV test uptake is promising for developing couple-based testing programs or re-instating premarital syphilis screening in China.
The large portion of physician-level variance supports that multilevel modeling is an effective tool for interpreting clinic-based HIV test uptake compared with either ignoring clustering or using marginal models. The lack of measured physician-level factors and patient-physician interactions in explaining HIV test uptake suggests the need for further research in this field. A more comprehensive assessment of physician attitudes to testing and testing-related variables could help explain the large portion of physician-level variance. Although previous research has shown that Chinese physicians have substantial HIV stigma,19,37 at least within urban developed areas such as the Pearl River Delta where this study took place, the level of HIV stigma among STI physicians seems unlikely to influence the expansion of HIV testing. Furthermore, physician HIV/STI training, which has been found deficient in other Chinese studies,38,39 seems to be sufficient for rolling out integrated syphilis/HIV testing in this context.
Integration of syphilis/HIV testing also offers a potential opportunity to quantify sexual risk in a subsection of individuals who refuse HIV testing. The high number of HIV test refusers with a positive rapid syphilis test (4 out of 8, data not shown) underscore the need for more studies of those who decline HIV testing. Improved patient counseling and physician training on the frequency of HIV testing and counseling for married STI patients may improve HIV test uptake.
This study has several limitations. First, the study was done at 6 STI clinics in the Pearl River Delta and so may not be representative of STI clinics in other parts of the province or region. Second, self-reporting bias regarding same-sex behaviors precludes a more formal assessment of men who have sex with men in this sample, similar to other studies in STI clinics.36 Finally, this study was done as part of a research project and so ordinary clinical practice may result in lower uptake rates, consistent with our finding that the HIV test uptake rate slightly decreased after 3 months of the program.
This pilot supports the utility of integrated syphilis/HIV testing, suggesting the importance of integrated syphilis and HIV programs. Many individuals at increased risk of both syphilis and HIV infection are evaluated at STI clinics in China, and care should be taken to evaluate those who decline HIV testing. The Chinese Ministry of Health recently released a 10-year syphilis prevention and control plan that establishes explicit goals for syphilis/HIV integration.40 This political commitment should help translate research findings from this study into specific guidelines and practice likely to find substantial numbers of patients with unrecognized HIV infection.
We would like to thank all the participants and members of the Plum Blossom team who made this possible: Dr Tinglu Ye, Dr Xuqi Ren, Dr He-Kun Lu, Dr Bao-Yuan Zhang, Dr Shu-Jie Huang, Dr Xue-Ling Tan, Dr Wei-Jun Deng, Dr Jian-Xin Yu, Dr Yun Feng, Dr Jing-Feng Huang, Dr Xiao-Xiong Huang, Dr Hua Peng, Dr Sen-Miao Zhang, Dr Fang-Mei Chen, and Dr Xiao-xi Zhang. Thanks to Professor Baoxi Wang (National STD Control Center), Dr Heather Ribaudo (Harvard), Professor Gail Henderson (UNC Chapel Hill), Professor Sten Vermund (Vanderbilt), Professor Arthur Kleinman (Harvard), Dr Rochelle Walensky (Harvard), Professor Martin K. Whyte (Harvard), Professor Rosanna Peeling (London School of Hygiene and Tropical Medicine), Professor James Maguire (Harvard), and Dr Sarah Hawkes (University College London).
1. Chen ZQ, Zhang GC, Gong XD, et al. Syphilis in China: results of a national surveillance programme. Lancet
2. Tucker JD, Chen XS, Peeling RW. Syphilis and social upheaval in China. N Engl J Med
3. Lu L, Jia M, Ma Y, et al. The changing face of HIV in China. Nature
4. HIV Infection—Guangdong Province, China, 1997-2007. MMWR Morb Mortal Wkly Rep
5. Powers KA, Poole C, Pettifor AE, et al. Rethinking the heterosexual infectivity of HIV-1: a systematic review and meta-analysis. Lancet Infect Dis
6. Rou K, Sullivan SG, Liu P, et al. Scaling up prevention programmes to reduce the sexual transmission of HIV in China. Int J Epidemiol
. 2010;39(suppl 2):ii38-ii46.
7. Ministry of Health of the People's Republic of China. China 2010 UNGASS Country Progress Report
. Beijing, China; 2010.
8. Ma W, Detels R, Feng Y, et al. Acceptance of and barriers to voluntary HIV counselling and testing among adults in Guizhou province, China. AIDS
. 2007;21(suppl 8):S129-S135.
10. Chen S-M, Liu D-C, Zhang FR. STD clinic provider-initiated HIV testing and counseling analysis. Paper presented at: 5th National STD Conference; May 2010; Haikou, China.
11. Bassett IV, Walensky RP. Integrating HIV screening into routine health care in resource-limited settings. Clin Infect Dis
. 2010;50(suppl 3):S77-S84.
12. NCSTD/CCDC. 2007 STD Epidemiology. In: Epidemiology
13. Yang LG, Tucker JD, Yang B, et al. Primary syphilis cases in Guangdong Province 1995-2008: opportunities for linking syphilis control and regional development. BMC Public Health
14. Tan N, Messina JP, Yang L, et al. A spatial analysis of county-level variation in syphilis and gonorrhea in Guangdong Province, China. PloS ONE
15. Yang LG. Annual STD Surveillance for Guangdong Province, China. Presented at: China National STI Control Conference; May 15, 2010; Haikou, China.
16. Tucker JD, Yang LG, Zhu ZJ, et al. Integrated syphilis/HIV screening in China: a qualitative analysis. BMC Health Serv Res
17. Anand S, Fan VY, Zhang J, et al. China's human resources for health: quantity, quality, and distribution. Lancet
18. Arbelaez C, Wright EA, Losina E, et al. Emergency provider attitudes and barriers to universal HIV testing in the emergency department. J Emerg Med
. 2009. [Epub ahead of print].
19. Li L, Wu Z, Zhao Y, Lin C, et al. Using case vignettes to measure HIV-related stigma among health professionals in China. Int J Epidemiol
20. Parish WL, Laumann EO, Cohen MS, et al. Population-based study of chlamydial infection in China: a hidden epidemic. JAMA
21. Conover WJ. Practical Non-Parametric Statistics
. 2nd ed. New York, NY: John Wiley and Sons; 1980.
22. Goldstein H. Multilevel Sstatistical Models
. 3rd ed. London, United Kingdom: E. Arnold Oxford University Press Distributor; 2003.
23. Raudenbush SW, Chan WS. Application of a hierarchical linear model to the study of adolescent deviance in an overlapping cohort design. J Consult Clin Psychol
24. Subramanian SV, Jones K, Kaddour A, et al. Revisiting Robinson: the perils of individualistic and ecologic fallacy. Int J Epidemiol
. 2009;38:342-360; Author reply: 370-343.
25. Browne WJ. MCMC Estimation in MLwiN (Version 2.10) Centre for Multilevel Modelling
Bristol, United Kingdom: University of Bristol; 2009.
26. Diez Roux AV, Aiello AE. Multilevel analysis of infectious diseases. J Infect Dis
. 2005;191(suppl 1):S25-S33.
27. Dickinson LM, Basu A. Multilevel modeling and practice-based research. Ann Fam Med
. 2005;3(suppl 1):S52-S60.
28. Wang Y, Li B, Zheng J, et al. Factors related to female sex workers' willingness to utilize VCT service: a qualitative study in Jinan City, Northern China. AIDS Behav
29. Wang B, Li X, Stanton B, et al. Correlates of HIV/STD testing and willingness to test among rural-to-urban migrants in China. AIDS Behav
30. Wang XW, Liu Y, Dong BQ, et al. Provider-initiated testing and counselling for human immunodeficiency virus among tuberculosis patients in Guangxi. Int J Tuberc Lung Dis
31. Zhou Z, Meyers K, Qian H, et al. Integrating HIV, hepatitis B, and syphilis screening and prevention in antenatal care and labor and delivery services in Yunnan Province, China. Presented at: XVIII International AIDS Conference; July 23, 2010; Vienna, Austria.
32. Kranzer K, McGrath N, Saul J, et al. Individual, household and community factors associated with HIV test refusal in rural Malawi. Trop Med Int Health
33. Freeman AE, Sattin RW, Miller KM, et al. Acceptance of rapid HIV screening in a southeastern emergency department. Acad Emerg Med
34. Wong FY, Huang ZJ, Wang W, et al. STIs and HIV among men having sex with men in China: a ticking time bomb? AIDS Educ Prev
35. Wu Z, Sun X, Sullivan SG, et al. Public health. HIV testing in China. Science
36. Cheng JQ, Zhou H, Hong FC, et al. Syphilis screening and intervention in 500,000 pregnant women in Shenzhen, the People's Republic of China. Sex Transm Infect
37. Hesketh T, Duo L, Li H, et al. Attitudes to HIV and HIV testing in high prevalence areas of China: informing the introduction of voluntary counselling and testing programmes. Sex Transm Infect
38. Zhou H, Chen XS, Hong FC, et al. Risk factors for syphilis infection among pregnant women: results of a case-control study in Shenzhen, China. Sex Transm Infect
39. Zhou P, Qian Y, Xu J, et al. Occurrence of congenital syphilis after maternal treatment with azithromycin during pregnancy. Sex Transm Dis
40. Tucker JD, Cohen MS. China's syphilis epidemic: epidemiology, proximate determinants of spread, and control responses. Curr Opin Infect Dis
Keywords:© 2011 Lippincott Williams & Wilkins, Inc.
China; HIV; multilevel modeling; syphilis; STI clinic; testing